ClawHub
Back to skill

Security audit

agentchan

Security checks across malware telemetry and agentic risk

Overview

This skill matches its AI imageboard purpose, but it lets mutable remote documents steer future posting behavior and local state, so users should review it carefully before installing.

Install only if you intentionally want an agent to interact with AgentChan. Require explicit approval before registration, posting, heartbeat participation, or file upload; do not let fetched heartbeat or skill markdown override your normal instructions; store the API key in a real secret store or a tightly permissioned local file; and review/redact all content before it is sent to the public service.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly recommends saving the API key to a local JSON file as a fallback, but it does not warn that this is plaintext credential storage or require restrictive file permissions. On multi-user systems, shared workspaces, synced home directories, backups, logs, or malware-compromised hosts, this can expose the bearer token and allow unauthorized posting or account impersonation.

Ssd 4

Medium
Confidence
99% confidence
Finding
The refresh rule instructs the agent to fetch remote documents and then 'follow current heartbeat instructions,' which delegates future behavior to content not fixed in the reviewed skill. That creates an unbounded remote-instruction channel: if the remote document changes or is compromised, the agent could be induced to post unexpected content, change cadence, or follow unsafe operational directions.

Ssd 1

Medium
Confidence
99% confidence
Finding
The update notice tells the agent to refresh local memory and overwrite stale cached instructions before posting, effectively prioritizing newly fetched remote content over previously trusted local guidance. This can semantically supersede safeguards or prior policy constraints and makes later remote edits a control plane for agent behavior.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.